[rtl/core] fixing SLINK IRQ bug #199

[stnolting]

This PR aims to fix the bug in SLINK's interrupt signal logic from #199.

@LeFl0w

[stnolting]

@LeFl0w You're ok with the modifications?

[LeFl0w]

I'm not sure about changes for lines 263 and 277.
You come from an iteration over only the implemented Slink bits to an iteration over every possible links numbers (even unimplemented ones).
This will produce dead code in case of SLINK_NUM_ not equal to 8.

I think you could keep the previous code , as your vector is initialized entirely to a default value with rx_tmp_v := (others => '0') and then you iterate over only implemented slink bits.

With your new code I 'll have to analyze how irq_rx_mode(i), rx_fifo_half(i) and rx_fifo_avail(i) behave when i is not referring to an implemented slink number. In my point of view it is more complicated to analyze without simulation.

[stnolting]

I'm not sure about changes for lines 263 and 277. This will produce dead code in case of SLINK_NUM_ not equal to 8.

That's right, but the synthesis tool should trim these signals.

I noticed that the MSB-aligned bits of irq_rx_en and irq_tx_en might not be initialized (not written in the bus write access process) if less than 8 links are implemented. That's where those chances came from.

However, feel free to update this PR with another proposal 😉

[stnolting]

This is obsolete now. See #202.